So, new year, and entry regulations are already with us and it’s almost time to start making plans again for this years car fun. For those new, I bought a Westfield with a turbo charged hayabusa engine, it should have been quick but we had a fairly dismal time on both the hillclimbs and sprints with it, actually getting worse over the few events we did last year. Below is an update on what has happened since. I am not ready yet to call this years plans but it’s getting closer.
The problem at the last event turned out to be a simple boost leak, but what it did pin point was that in the transition range from normal to boosted running we had an engine management issue of some type. The engine was very stuttery during the transition although fine before it & fairly fine after it. The boost leak just meant we were spending more time in this transition. The stutter was not a new problem, the engine has always been like this but each time we have traced a different issue that could be the cause and clearly none of them had been.
The good news it turned out was I had a lot of good logging from the running with the lambda sensor turned on. What I didn’t have though was a decent way to analyse the data. The DTA ECU software does not really have a log analysis package although there is a third-party application called E-Race that is OK, but not ideal. So, I broke out GWT and knocked up my own analysis code to plot the air-fuel ratio readings from the lambda sensor against the RPM and manifold pressures at a pretty fine granularity.
This showed the problem pretty directly, the fuel map was good but it wasn’t granular enough. In the transition between on & off boost (between 90kpa and say 125kpa, where 100kpa=atomospheric) the fuel needs rise very quickly, so fueling was good at 90kpa, good at 100kpa, and even good a 125kpa, but poor at 105,110 & 115kpa. The issue here is that the ECU interpolates between the fuel columns that are each at a fixed kpa value, but of course it does that linearly, and linear is not at all what the engine needs in this area.
The fix for this is fairly straight forward once you know it is there, you adjust the position of the fuel columns so there are more in the non-linear areas and fewer in the linear ones to improve the accuracy. Of course I had set the map up in a non-linear way already based on others advise of how to do for turbo engines, but the column positioning clearly wasn’t right for what this engine needed. We should have caught this doing dyno testing but you skip over this area very quickly in a straight power run so don’t notice the fueling going bad, it only shows itself on slow corners (and of course when you have a boost leak). Anyway, some spreadsheet calculations later I now had a new fuel map for the ECU ready for testing and so took the car off for a day at Barkston Heath.
The engine was now running so much better you would not have believed it was the same setup as before. On a couple of runs I thought the stutter might be back, each run was when running low on fuel and the stutter didn’t return after a top up. So problem #2 found, there is a minor starvation issue somewhere in the fuel system. I haven’t fixed this yet but its easy to avoid so it can wait.
Snow, and other distractions stopped me taking the car out for more testing but problem #3 was gifted via an email conversation about possible upgrades for the turbo with Holeshot (the original engine builders). The dyno shows the engine making ~270bhp at near full boost, but the claim is it should be making ~300bhp at this level, I had dismissed that as marketing BS but it was suggested I double checked the timing, so armed with a new timing gun I did and found the DTA running a few degrees retarded. This is a setting inherited from the default Hayabusa map for the DTA, I should have checked this sooner, not the first DTA default map setting that has caused problems, mea culpa.
Next test day is 30th Jan back at Barkston Heath. The first day there had shown that turbo lag (a delay between opening the throttle and making boost caused by the turbo needing to spin up) is maybe going to be an issue. With the odd engine running I had not really felt this much before but Barkston has a couple of very tight corners that really highlighted the issue now the fueling was much better. The retarded timing issue may well have been making this worse, but we will have to see on that one. In case this really turns into a problem I have configured things to allow testing a mild anti-lag (or bang-bang) setup. This is a fairly complex setup, the long and short is we can get the turbo generating boost when running round slow corners so that there is no lag when you open the throttle again. The downside is reduced turbo life, but maybe I can live with that on a competition car.
That’s been my winter of fettling, as always I would have liked to have done more but time is always more limited than you would like. Plans for this year I think are going to depend rather a lot on how things go at Barkston next time out …
